Admin's note: Paul is new to DV Info Net and
therefore may be unaware that this subject has
been extensively discussed here months ago
when the XL2 was first released.

--------------------------------------------------------------
I went to the canon xl2 camera overview at DVExpo WEST last week, and learned much valuable information. It's probably already been declared, but the canon xl2 does NOT have native 16:9 CCDs!
--------------------------------------------------------------

Admin's note: Actually yes they ARE native 16:9 CCD's:
for a complete graphical description of what's really going
on, please read my XL2 Watchdog article, "Canon XL2 CCD Block Overview (http://www.dvinfo.net/canonxl2/articles/article06.php)."

--------------------------------------------------------------
If you take the lens off and look at the CCD, you'll find it's shaped like a 4:3 CCD.
--------------------------------------------------------------

Admin's note: Actually there are three CCD's; if you
take the lens off and look at the camera head, you're
seeing a prism block on which all three are mounted.
The shape of each CCD is completely irrelevant -- they
could be circular, triangular, oblong, trapezoidal, 4:3,
perfectly square, 16:9, star-shaped, macaroni-shaped,
hexagonal, octagonal. All that really matters is the
TARGET AREA of the CCD, and in this case, the target
area of each CCD in the XL2 is in fact 16:9.

--------------------------------------------------------------
Canon chose to go with 4:3 ccds in order to make it's price point of under 5k.
--------------------------------------------------------------

Admin's note: Incorrect. they choose to go with the
4:3 shape because that's all they could buy (Canon
must buy the CCD's from suppliers as they do not
make their own, and at the time, all 1/3rd-inch size
CCD's are made in a 4:3 shape).

--------------------------------------------------------------
They simply have turned off the top and bottom of the ccd, to get a 16:9 signal out of it.
--------------------------------------------------------------

Admin's note: this produces a
native 16:9 image at 960 x 480 pixels.

--------------------------------------------------------------
To get 4:3 out of the ccd, they mask off the sides of the already smaller area!
--------------------------------------------------------------

Admin's note: this produces a
native 4:3 image at 720 x 480 pixels.

--------------------------------------------------------------
The skinny is if you shoot in 4:3 on an XL2, you're using MUCH less of a 1/3" CCD than on the XL1, and as such, you'll get a much deeper depth of field!
--------------------------------------------------------------

Admin's note: the difference in DOF really
is not that remarkable -- as the following
discussion reveals.

--------------------------------------------------------------
Just thought I'd give the heads up.
--------------------------------------------------------------

Admin's note: just thought I'd clear up
these misconceptions. Thanks -- Chris Hurd.

I'm not sure how this is really news - this has been detailed quite nicely already and in much greater detail. See the following:

http://www.dvinfo.net/canonxl2/articles/article06.php

What matters rather more than the size of the CCD block is the resolution and bit-depth which it samples video at. I think you'll find the XL-2 is actually rather a nice camera all things considered.

My point was to address posts like this one:
http://www.dvinfo.net/conf/showthread.php?s=&threadid=36277

Which wonders why the XL2 does not perform as well in low light situations (the smaller pixel size on the CCD results in less light being picked up) as well as letting people know the depth of field will probably be deeper on the XL2 than all other 1/3" ccds

That's certainly true, Paul, but I guess most people (me too) will mainly use the (true!) 16:9 mode of the XL2 which means, you use a larger CCD area than in 4:3 mode.

However, you'll never have a really shallow depth of field with such small CCDs. Even the 2/3" chips of ENG cams leave somewhat to be desired in this respect.

First: what does DoF have to do with whether the chips are 4:3
or 16:9 (yes I know the DoF increases in 4:3 mode) versues 2/3"
ENG camera's? There is no pro-sumer camera in this price range
with 2/3", simple.

Second: I'm going to go against this, the XL2 has true 16:9.

Paul, what would be your definition of true 16:9?

Mine is:In true 16:9 mode your field-of-view (16:9) will increase compared to 4:3 mode and the camera will use more pixels to sample your imageWhich is exactly what the XL2 and no other (non HDV) pro-sumer
camera in this price range does!

What if canon had chopped of the top and bottom portion of the
*CURRENT* CCD chips to get them into the 16:9 shape, would
that be true 16:9? (that's what the big boys basically do and that
is still the exact same procedure as what the XL2 does).

Canon just got a 4:3 sized CCD since it is CHEAPER.
If I'm not mistaken one of the HDV camera's does the exact same
thing.

Basically there are three ways (chip wise) to get 16:9

1. use a normal 4:3 chip and do a 16:9 stretch in post (loosing resolution and not increasing field-of-view)

2. use a normal 4:3 chip and get a 16:9 rectangle for it. The 4:3 rectangle is *INSIDE* this 16:9 rectangle and uses LESS resolution (this is what the XL2 does)

3. get a 16:9 shaped CCD which does the exact same thing as option 2

All in all the camera really does have true 16:9. And if you are
concerned about resolution. The camera has more resolution in
4:3 mode than most other camera's out there and this only
increases in 16:9 mode, so no concerns there either.

I can't wait to get my XL2!!

CCD block size has absolutely nothing to do with DOF. By definition, any CCD block which records in 16:9 is native, regardless of how much of the CCD block is used, or not used. Non-native, refers to obtaining 16:9 thru the use of letterboxing or an anamorphic lens.

The smaller the target in any imaging system, the greater the depth of field. Super 8 film has much greater depth of field than 35mm. DV has much greater depth of field than video cameras with larger chips.

The business about the XL2 having greater depth of field in the 4x3 mode is not quite right. To achieve 4x3 the XL2 simply lops off the pixels on the sides. The optics don't change. All the reflected light streaming in through the lens is washing over the whole chip. It's just that the pixels left in the middle display a 4x3 image. Shooting in 4x3 on the XL2 simply means that you're recording less information from the chip.

Many folks find the depth of field in DV excessive. The dinkiest little DV camera you can buy will have more depth of field than you can shake a stick at. The problem is that all that information can become visual clutter...you know...it is harder to get the viewer's eye to go where you want it to.

The point of this is to say that expansive depth of field is something you get by the yard in DV. Finding ways of getting less of it, like using the 20X lens on the XL2, is really nice to know about.

I don't think DOF is related to target size, but, rather, the aperture of the optic that was designed for a particular target size. A lens designed for a smaller target size will be smaller and have higher DOF as a result of the smaller max aperture. Note that if you keep the f/number the same, there is no change in DOF.

If what you're saying is right, then the center of a lens would have more DOF than the outer edges. I don't think this is correct.

What about 1/3 inch CCD's having more DoF than 35mm film?

Just as the Target Size can vary as in the above paragraph, so can the size of the film stock or the size of the CCD image sensor. If the Target Size in front of the lens is to remain the same when we change CCD sizes, then DoF will indeed change.Source: The Ultimate Depth-of-Field Skinny (http://www.dvinfo.net/articles/optics/dofskinny.php)

The article referenced by Rob Lohman, above, states the following:

"If all formula variables stay the same and the Target Size (CCD) behind the lens changes, then DoF will not change. If the Target Size changes in front of the lens (by changing focal length or len-to-subject distance), then DoF will change."

In the context of this article, "target size" refers not to the detector size, but, to the image of the subject, i.e. target. That's quite different than the detector size.

Are you saying this to not confuse things with your posts Bill or
are you disagreeing that sensor size doesn't matter (compared
to other "sensor" sizes)?

LOL...
No, I really don't want to confuse things. This particular issue has been extremely confusing to me, as Jeff Donald will attest. I finally got it straight, tho'. I agree, Rob, sensor size does NOT matter. Image size on the sensor matters.

Well, yeah, but that's basically the same thing since the image size
is targeted to the size of the sensor.

But this also seems to be true if you mount a EOS lens and get
the 7.2 magnification, you still don't get a shallower DoF for
example.

<<<-- Originally posted by Bill Ravens : I don't think DOF is related to target size, but, rather, the aperture of the optic that was designed for a particular target size. A lens designed for a smaller target size will be smaller and have higher DOF as a result of the smaller max aperture. Note that if you keep the f/number the same, there is no change in DOF.

If what you're saying is right, then the center of a lens would have more DOF than the outer edges. I don't think this is correct. -->>>

If it's the lens that determines the DOF, why doesn't putting a 35mm lens with a mechanical adapter in front of the XL2 decrease DOF?

Just trying to understand the technicalities of it. I want to know if I'll be able to find work arounds in the future. I'm currently shooting my first video movie (did a few on film before that) and the deep DOF is driving me crazy. Just to do a pull focus from an object to an actor sitting at a table, I had to cheat with distances by using a tele focal length in order to crush all the planes together, and then put the object in front of the camera on a stool, several feets from where it should actually be in the scene normally. It works somewhat, but it's a real pain.

Can't wait to try a Mini35 adapter. Only problem for me is the price. But if I could decrease the DOF by using a normal PL to XL adapter and rent a few 16mm cine lenses, I would do it in a heart beat.

David Lach wrote:
"If it's the lens that determines the DOF, why doesn't putting a 35mm lens with a mechanical adapter in front of the XL2 decrease DOF?"

The answer goes back to the image size part of this equation. With an EOS lens, the magnification factor forces you to mave further away from the subject(target) in order to make it appear the same size on the sensor as with the standard lens(IOW you don't want to overfill the image frame). As soon as you move further away, the DOF increases so that whatever you gained by going to a 35mm lens, you lose by having to go further away. In the end the DOF stays the same.

The reason the mini-35 gives less DOF is because you're imaging a 2D ground glass image at a fixed distance from the lens, without changing the lens to subject(target) distance.

I you set a subject (like somebody's head) focussed, and you want a it at certain size relative to the total picture height (or width), then DOF will be shallower with larger "sensors" (CCD, film, ground glass..), and the total DOF range (near + far distance) will even be constant over a wide range of zoom settings.

Depth of field depends on lens focal length, aperture, focus distance and the acceptable diameter of the circle of confusion. Period. But the smaller the sensor is the shorter the focal length needed to cover it and the smaller the tolerable circle of confusion diameter so that the depth of field (which depends roughly on the square of the focal length and the first power of the circle of confusion diameter) will depend on the sensor size.

The hyperfocal distance is simply (focal-length)/(f-stop*normalized_diameter) with normalized_diameter being the diameter of the circle of confusion nomalized by the focal length. This number is approximately constant (because we want equal final image clarity at viewing size irrespective of the size of the film or sensor which recorded the image). The maximum depth of field is half the hyperfocal distance to infinity when the lens is focused at the hyperfocal distance. At other focus distances the dependence isn't quite so simple but the general principal remains the same.

By definition, any CCD block which records in 16:9 is native, regardless of how much of the CCD block is used, or not used. Non-native, refers to obtaining 16:9 thru the use of letterboxing or an anamorphic lens.
Or digital stretching.

Just a quick follow-up, while I warmly welcome all new members to DV Info Net, I would ask for some basic research before stepping up to make sweeping technical assumptions in an open forum, especially when they turn out to be so inaccurate.

Nothing personal to Paul Gnuyen, and not to reflect badly upon him, but I did feel compelled to insert some corrections to his original post (the content of which has been left intact) in order to prevent the casual reader from getting the wrong idea about what's really going on with the XL2's CCD block.

Basically, the gist of it is this:

The shape of each CCD is completely irrelevant -- they could be circular, triangular, oblong, trapezoidal, 4:3, perfectly square, 16:9, star-shaped, macaroni-shaped, hexagonal, octagonal, or some other shape entirely. All that really matters is the TARGET AREA of the CCD, and in this case, the target area of each CCD in the XL2 is in fact 16:9.

A few points:
--------------------------------------------------------------
Canon chose to go with 4:3 ccds in order to make it's price point of under 5k.
--------------------------------------------------------------

Admin's note: Incorrect. they choose to go with the
4:3 shape because that's all they could buy (Canon
must buy the CCD's from suppliers as they do not
make their own, and at the time, all 1/3rd-inch size
CCD's are made in a 4:3 shape).

David Castillo, Senior Technical Representative for Canon Video said in two sessions the reason the 4:3 CCDs were chosen were to keep the price point under 5k and going with NATIVE 16:9 CCDs would have brought it over 7k. He used this terminology of NATIVE 16:9 CCDs.

Admin's note: Actually yes they ARE native 16:9 CCD's:
for a complete graphical description of what's really going
on, please read my XL2 Watchdog article, "Canon XL2 CCD Block Overview."

As I said earlier, it was specifically said by a canon representative who was on the Xl2 design team, that the Canon XL2 does not have native 16:9 CCDs. Maybe this forum uses a different definition of native, but I'm just reporting what was said at this canon sponsored event by canon people.

--------------------------------------------------------------
The skinny is if you shoot in 4:3 on an XL2, you're using MUCH less of a 1/3" CCD than on the XL1, and as such, you'll get a much deeper depth of field!
--------------------------------------------------------------

Admin's note: the difference in DOF really
is not that remarkable -- as the following
discussion reveals.

Whether it is remarkable or not, it is an issue and was discussed at the event. You will get deeper depth of field using the same lenses shooting in 4:3 on an xl1 or xl1s than an xl2.

err that should be "more shallow depth of field"

Like Chris wrote, "native CCD" behavior doesn't relate to the form factor of the CCD sensor. It only means that the readout architecture of the CCD is representative for the system's vertical line number (480/576) interlaced or progressive. So XL2 has a native 16:9 behavier.

Okay, let's get this sorted out once and for all (I hope).

1. yes, the CCD's are not "native 16:9" as in the shape of the CCD's are NOT 16:9. Canon used NATIVE here to indicate the size of the CCD's, not the output signal. So let's use the word native for CCD size, and "true" whether we get the increase resolution / FoV.

2. as Chris indicated the shape of the CCD's has no bearing on whether the signal generated is a true 16:9 anamorphic signal.

3. keep in mind that an anamorphic attachment creates a true 16:9 signal as well, even if the CCD's/camera does not support it (that's the whole idea).

4. the definition of true 16:9 (as in getting a benefit) is an increased resolution in sampling and an increase field of view (fov). Contrary to popular believe, within the DV world true 16:9 does not increase the horizontal resolution (it does increase the horizontal sampling), but the vertical

5. vertical resolution is increase with true 16:9 in DV since the signal is still downsampled to 720x480/720x576. This yields a vertical resolution increase since you get the full 480/576 lines instead an upsampled one

In the end the size of the chips does not matter at all. What
matters is how they get the 16:9 signal and there are three ways:

1. an electronic stretch: this is what the cheaper/most camera's do and you loose resolution

2. using a larger area on the CCD: this is what the XL2 does and increases resolution and field of view (again, CCD shape/size has nothing to do with this)

3. using an anamorphic attachment: this will also increase resolution and field of view

So the only two methods that are true 16:9 are method 2 & 3.

To recap: yes the CCD's in the XL2 are not native 16:9 (ie, the
shape of the CCD's is 4:3), but the signal generated is full 16:9
anamorphic so it is "true" 16:9.

Sorry Rob, but even "native CCD's"(in your definition) only have a 16:9 readout area (just like XL2) but the global CCD structure has overhead pixels ( just like all CCD's have). So, a 16:9 shaped CCD doesn't exist. Only a 16:9 shaped readout area CCD does exist

Andre: that would be even better! But I could've sworn that
companies claim they are using native 16:9 (shaped) CCD's....

Ofcourse it still doesn't change anything in the story (as you know).

Hi Paul,

David Castillo, Senior Technical Representative for Canon Video said in two sessions the reason the 4:3 CCDs were chosen were to keep the price point under 5k and going with NATIVE 16:9 CCDs would have brought it over 7k. He used this terminology of NATIVE 16:9 CCDs.

David and I were telling you the same thing. I stated that 1/3rd-inch CCD's did not exist in the 16:9 shape (they would have to be developed). He stated it would have been too expensive (they would have to be developed). Both of us are saying there was no such thing as a 1/3rd-inch CCD in the 16:9 shape available for the XL2. My take was that Canon would need to have them OEM'ed. David's take was that they could have had them made but it would have been too expensive. Either way it's the same thing: there was no such thing as a 1/3rd-inch CCD in the 16:9 shape available at the time for the XL2.


As I said earlier, it was specifically said by a canon representative who was on the Xl2 design team, that the Canon XL2 does not have native 16:9 CCDs. Maybe this forum uses a different definition of native, but I'm just reporting what was said at this canon sponsored event by canon people.

Well, it certainly is no secret that Canon USA's marketing division is quite frequently its own worst enemy. You've just provided a perfect example of this problem. Fortunately for Canon, there are people like myself and many others on this board who at least attempt some degree of diligence by following in the wake they create and trying to repair their own self-inflicted damage. This thankless job is done for the sole purpose of providing our readers with the most accurate technical information possible, in order to aid their selection and proper use of the tools of the DV trade. Sadly there are occasions where there is a conflict between what we tell you and what a manufacturer tells you. That doesn't happen very often, but when it does, it's definitely in your best interest to believe us over them.

Canon USA does not make the XL2. The camera is made in Japan. Canon USA does the marketing and selling for a chunk of North America, and sometimes they know what they've got and sometimes they don't. It's really too bad that I can't be around to rescue them from every instance of bad semantics, but what you were told with regard to "not having native 16:9 chips" had more to do with their SHAPE than anything else. Of course the XL2 produces native 16:9 digital video. And as I've previously stated, the shape of each CCD is completely irrelelevant to its actual target area. It could have triscuit-shaped CCD's but still give you native 16:9 from its actual target area. It could have 8" by 10" size CCD's but still give you an actual target area of only 1/4 inch.

You will get deeper depth of field using the same lenses shooting in 4:3 on an xl1 or xl1s than an xl2.

Paul, I understand that the original intent of your post was to point out that the smaller physical size of the XL2 CCD's target area yields the equivalent of a 1/4-inch CCD when shooting in 4:3 mode. You are correct about that and there is no argument here. What I'm taking issue with is, why complain about the lack of a shallow DOF with 1/4-inch chips when that lack is already present at 1/3rd-inch. In other words, my point is that you're really not gaining much DOF control by going from 1/4-inch to 1/3rd-inch. Neither size is good for achieving shallow depth of field. If you want shallow DOF, you need to use the proper tools to create shallow DOF. In the case of the XL2, that would involve the rental of a P+S Technik Mini-35 converter and the appropriate choice of 35mm motion picture lenses. Now there's your shallow depth of field.

Regarding "native 16:9" and all the confusion and all the definitions...

May I humbly submit that "a difference that makes no difference, IS no difference"? I mean, who cares whether it's technically this or technically that, as long as the picture being delivered to the frame is full resolution?

The XL2 delivers full resolution. So does the PDX10. So does the FX1. The XL2 and PDX10 use 4:3 CCD's, the FX1 uses 16:9 CCD's... but who cares? You're still getting the full resolution, and that should be the determining factor: do you get the full resolution of the video frame?

On a camera like the XL1 or VX2000 or whatever, you don't get the full resolution, you get an electronically-stretched image that doesn't provide full resolution. Let's make that the dividing line: do you get full resolution or not? If you do, it's "true" or "native" 16:9. If you don't, it's "fake" 16:9.

If you want to pick on the XL2, say it's a 1/3" camera in 16:9 mode and a 1/4" camera in 4:3 mode. But that has nothing to do with the "native" or "true" 16:9 discussion.

Barry, what do you meam by "full resolution"? I first suppose we are talking about vertical resolution. Further on, do you mean 480/576 lines with active video? Then also uprezzed pics have ""full resolution". In all cases whether it's uprezzed video or native, there is vertical aliasing and vertical resolution limitation. The only difference is that that besides the optical aliasing (straddling...) and resolution limitations you get additional aliasing from the in camera uprez. So I thinkyour "full resolution" is as fuzzy as the "native" approach.

Just call it 960 by 480. Nothing fuzzy with that, right?

Yeah, everything gets fuzzy. And calling it 960x480 confuses issues too, because by the time it hits tape it's 720x480. And then they wonder where that went, and isn't the FX1 better because it's 960x1080 and... on and on...

For "full resolution" I was referring to shooting a resolution chart. If the camera will spit out a resolution chart that shows 540 lines of resolution, then that's what I was calling "full resolution" -- which is the limit of DV's recording capability. But, as you point out, that's not the be-all-end-all definition, either...

Why can't we just define it like:

" if the field-of-view widens when you switch from 4:3 to 16:9
mode your camera (or anamorphic attachment) does true 16:9
and you get an increase in resolution. Otherwise it is a fake 16:9 "

That should be precise and easy to explain.... no?

Seems reasonable, but...

... I don't know, it just seems to legitimize the compromise Canon made for 4:3 mode. I mean, why doesn't the XL2 use its full 4:3 CCD when in 4:3 mode? Primarily it would appear that they did it for marketing purposes: to make you think you're getting "more" when you go to 16:9 mode, as opposed to what's actually happening, which is that you're getting "less" when you go to 4:3 mode.

Look at it this way: what if Canon did everything with the chip exactly as they did, but instead of using a 1/4" subset of the 16:9 patch, they let you use the full surface of the chip for 4:3 mode. By that token, you'd get the same "true" 16:9 you already have, which would fulfill my definition of "full resolution", and you'd have a much larger chip surface area for 4:3 (ostensibly better than what we have now, right?

But by your definition, it would fail the test of "true" 16:9, because the FOV wouldn't get wider in 16:9 mode.

<<<-- Originally posted by Barry Green :
Look at it this way: what if Canon did everything with the chip exactly as they did, but instead of using a 1/4" subset of the 16:9 patch, they let you use the full surface of the chip for 4:3 mode. By that token, you'd get the same "true" 16:9 you already have, which would fulfill my definition of "full resolution", and you'd have a much larger chip surface area for 4:3 (ostensibly better than what we have now, right?

-->>>

Then you would have a situation where the 4x3 image would have more pixels to sample from making it "higher res" than the 16x9 mode. In which case everyone would be up in arms again because 16x9 would be compromised. This would lead to a desire for an anamorphic adapter so that 16x9 freaks can take full advantage of all the pixels in 4x3 mode. It is a catch 22. At least the way it is now, Canon has clearly marketed the camera to succeed in 16x9 mode and be competetive in 4x3.

There are some technical reasons why the full CCD height readout would not be applied in 4:3. The first is optics related: a larger chip diagonal could lower the image quality in the corner area's (vignetting, aberations..). A second one would be the need for a different line readout algorithm (polyphase filter)for 4:3. A third reason could be that part of the CCD is optically screened off and used as a buffer memory in progressive mode (semi FT mode)

The active region in 4:3 is 720 x 480 - exactly the number of samples in each direction required for transfer to tape. Thus no interpolation is required. If the full sensor were (could) be used then subsampling would be required in which the extra resolution would be thrown away anyway (Note: in horizontal up by 3 and down by 4 and in vertical up by 2 and down by 3 gives 729 x 480 so polyphases wouldn't be necessary). What might be nice would be to do the antialiasing filter digitally and even perhaps to give the user some control over its cutoff so that he could eliminate jaggies (to some extent, at least) where they are annoying and go for full resolution where they aren't.

Agree with Andre that there might be lens coverage problems at the corners and also that those inactive areas are probably not available.

There can't be lens coverage issues -- the chip is exactly the same size as the XL1's, and the camera uses XL1 lenses and the XL1 can use XL2 lenses.

So in theory on a different camera that might apply, but for this camera it should be a complete non-issue.

True if the XL1s used the whole chip (i.e. no fallow pixels as on the XL2 set). Probably true for all practical purposes even if it didn't. The remark was really based on the observation that use of the whole chip for 4:3 on the XL2 would take you 12% further from the lens axis than the 16:9 mode.

Well, while all you silly persons are arguing semantics, I think I'll go shoot some video. ;-)~

Happy Holidays, everyone. Catch the spirit!

As said earlier, if they would do a full 4:3 readout we would be
back at the XL1(S) camera where we do not have a full 16:9
mode but a faux one. That is the whole idea. If you have a true
16:9 camera 4:3 mode WILL ALWAYS be lower resolution.

Heck, this is even true with an anamorphic attachment and a plain
4:3 camera. With 16:9 attachment on you get true 16:9 and an
increased resolution. Withouth it you are at plain 4:3 and you have
lost the increased resolution. This is always true!

So you have the following choices:

1. fake 16:9
- no resolution or FoV increase
- no resolution change between 4:3 & 16:9

2. true 16:9 (whatever method)
- resolution and FoV increase in 16:9 compared to 4:3
- 4:3 is lower resolution than 16:9 (that is the whole idea)

But as I've said time and time again on this board: I personally
don't care about resolution. The chance of my footage coming out
on some medium where people (on usually crappy equipment)
will actually see the difference is SMALL.

Yes we wan't the best quality we can get, but why not focus all
of this energy at making some good stories and movies?

Resolution is only a small factor in the succes of a movie, I can
garantuee you that!

<<<-- Originally posted by Rob Lohman :
2. true 16:9 (whatever method)
- resolution and FoV increase in 16:9 compared to 4:3
- 4:3 is lower resolution than 16:9 (that is the whole idea)
-->>>

Rob, if DV is 720x576 (or 720x480), whether it is 4:3 or 16:9, why do you say there is an increase in resolution in 16:9 compared with 4:3? I would agree there is an increase in resolution if you compare native 16:9 to stretched 16:9, but not that one set of 720x576 samples is higher resolution than another. Am I missing something?

Richard Hunter

<<<-- Originally posted by Richard Hunter : <<<-- Originally posted by Rob Lohman :
2. true 16:9 (whatever method)
- resolution and FoV increase in 16:9 compared to 4:3
- 4:3 is lower resolution than 16:9 (that is the whole idea)
-->>>

Rob, if DV is 720x576 (or 720x480), whether it is 4:3 or 16:9, why do you say there is an increase in resolution in 16:9 compared with 4:3? I would agree there is an increase in resolution if you compare native 16:9 to stretched 16:9, but not that one set of 720x576 samples is higher resolution than another. Am I missing something?

Richard Hunter -->>>

The concept is that the more pixels that are sampled from the CCD the higher the perceived resolution is, regardless of the actual "DV" pixel count. To give an over simplified example I take a high res photo taken from a 35mm still camera and drop it into a DV piece I am editing on my PC. When I cut from the actual DV camera footage to the 35mm shot there is a huge difference in the resolution and clarity of this shot. IT is obviously a much higher qulaity source image. This proves that regardless of the DV compression, it is possible to see more information if the source material is more detailed.

Therefore a 940x480 16x9 source image from the CCD on the XL2 has the potential to look more clear and precise than an image acquired from a 720x480 sample from a source CCD on other prosumer level cameras, regardless of the compression after that stage.

Make sense?

Hmm. Well said there, Marty!

Hi Marty, thanks for the explanation. OK, there are probably other good reasons why the image from 35mm film is clearer than those from a DV camera, but I get what you are saying. Even so, the 940x480 resolution would apply only for a system such as used in the XL2 and not to native 16:9 per se, right? For example, if the CCD is 16:9 aspect ratio and has 720x480 pixels, it would still be native 16:9 but I don't see that it would have higher resolution than 4:3 at 720x480.

Richard

Besides the excellent point made by Marty you also have an
actual increase in resolution: the vertical resolution. Let me
clarify by a short example:

Normal NTSC DV is 720 x 480 pixels.

A fake 16:9 camera that does an anamorphic stretch does the
following:

1. crop the signal to 720 x 364

2. stretch that back out to 720 x 480 (upsampling)

(or it can do the stretch first and then a crop, but that's the exact
same thing)

So you are now working with a 720 x 364 pixel image instead of
720 x 480! Which looses you 116 lines of resolution (which would
be the same if you letterboxed it for example)!

A true 16:9 camera like the Canon samples the image in 16:9
mode at 960 x 480: http://www.dvinfo.net/canonxl2/articles/article06.php

So the only thing this camera has to do is:

1. stretch the signal back to 720 x 480 (downsampling)

In this case you "loose" resolution in the HORIZONTAL (since you
go from 960 to 720 pixels per line). However, the VERTICAL can
stay the exact same.

If you compare this to the faux method you see that you gain the
116 lines of resolution that you lost there.

So you have two things to gain from a true 16:9 camera or a
camera with an anamorphic attachment that records in DV:

1. an increase in VERTICAL resolution

2. a higher sampling resolution which yields a richer image (Marty's point)

This is the exact same thing that happens with a DVD (the image
recorded there is 720 x 480 as well, anamorphic 16:9 or not!).
The increase is in vertical resolution and not horizontal or both.

Ofcourse the best thing would be to not re-sample/scale/stretch
the image at all and get the full 960 x 480 image, but that is just
not allowed in the DV standard (for bandwidth reasons at that
time etc.), so that's not going to happen.

The only thing one can do is go to a HD format for even greater
resolution, but even those sometimes have stretching.

For example, the new Sony camera records at 1440 x 1080 instead
of the full HD 1920 x 1080 signal. What they have done is that
they are recording at a different pixel aspect ratio so that 1440
must be horizontall stretched (upsampled) to 1920 which ofcourse
looses you some resolution. Although it is still (much) more than
960 x 480 in plain SD resolution.


Richard: to answer your question: with such an attachment the
chips will not change and you can argue about resolution. In this
case you will still benefit from my point #1 in the list above (since
you get to keep the full vertical resolution) and point #2 is the one
you can argue about. In my opinion you gain in optical resolution
instead of true pixels like a true 16:9 CCD.

The end result should however be the same with a good enough
piece of anamorphic glass in front of your camera!


I hope this post has clarified some things!

RE: "This proves that regardless of the DV compression, it is possible to see more information if the source material is more detailed."

This statement, at first blush, appears to violate the sampling theorem. A sensor with 720 pixels in its horizontal dimension cannot represent more than 360 cycles per picture width. A sensor with more than that (e.g. the 960 in the XL2) can capture higher spatial frequencies but they cannot be correctly displayed if the image is downsampled to 720. "Proper" downsampling would require that all frequencies above 360 CPW be attenuated below the visibility threshold. If this is not done then those frequencies will appear as aliases. "Proper" downsampling requires a "brick wall" filter i.e. one whose response goes from unity gain at 359.9999 CPW to 0 at 360.0001. Such a beast does not exist. Practical filters must start to roll off gain well below 360 in order to have sufficient attenuation above 360. The wider this so called transition band the easier the filter is to implement i.e. the fewer "taps" it requires or, put another way, the shorter its impulse response needs to be. Impulse response duration is important when working with images because of its effects on the edges of the picture.

So why does 35 mm downsampled to DVD look better than the stuff coming out of our XL2's? I believe that the answer is that in the telecine process the frames are substantially oversampled which allows the interpolation filter to have a narrower transition band. Thus we get closer to 360 CPW, the fundamental limit of the medium than we do with the filter implemented in the camera.

What about 4:3 where there is no downsampling and we can't blame the low pass filter? I'm afraid the answer there is that the camera, though its chips have 720 pixels, is not capable of anything near 360 CPW resolution. This could be caused by MTF limitations in the lens or other optical components (which doesn't bode well for an HD version of the XL series) or by the infamous Kell effect or combinations of the above. I find the manufacturers very disingenuous when it comes to discussions of the resolutions of these cameras (and not just Canon). There's lots more to it than pixel counts! Why won't they show us resolution targets or MTF curves?

To summarize: The absolute limit of the DV system is 360 cycles per picture width. The fact that telecine video looks sharper on DV than XL2 video means that DV is not the limiting factor in the XL2's resolution. So in essence I agree with the quoted statement up to somewhat less than the ultimate limit which DV does impose: 360 CPW.

Synchronious sampling a CCD structure (e.g 720 hor samples for 720 hor pix) is not a nyquist problem. And indeed higher pix counts need to be downsampled with all the low pass filtering problems and limits involved. The basic problem is the optical sampling by the physical CCD structure: the scene is being sampled by 720 (hor) photosensors( simular story is true in vert direction). If this scene contains spatial (horizontal) components which can interfere with the pixalated CCD structure we get optical aliasing which cannot be reduced by whatever filtering on the electrical signal. So optical lowpass filtering (preferable steep) is the only way to go. Consumer cams quite often only have the MTF and/or diffraction limits as optical low pass filtering. Not at all steep and thus introducing alising vs res limits as a trade off. If we got a cam with an infinite number of pix there would not be optical aliasing, and just an ideal (digital) lowpass filter would result in 'perfect' DV samples. The approach on the optical oversampling (a multitude of 720 hor pix) and steep filtering is what happens in 35 mm telecine processing., and this makes the DV pix better.

<<<-- Originally posted by Rob Lohman : Besides the excellent point made by Marty you also have an
actual increase in resolution: the vertical resolution. Let me
clarify by a short example:

Normal NTSC DV is 720 x 480 pixels.

A fake 16:9 camera that does an anamorphic stretch does the
following:

1. crop the signal to 720 x 364

2. stretch that back out to 720 x 480 (upsampling)

(or it can do the stretch first and then a crop, but that's the exact
same thing)

So you are now working with a 720 x 364 pixel image instead of
720 x 480! Which looses you 116 lines of resolution (which would
be the same if you letterboxed it for example)!

A true 16:9 camera like the Canon samples the image in 16:9
mode at 960 x 480: -->>>

Hi Rob. I agree that native 16:9 is higher resolution than stretched 16:9, but what I was questioning was whether it is higher than native 4:3, if both are sampled at 720x480. I still don't really see that it can be so. Maybe your point is that most cameras will not offer both native 16:9 AND native 4:3, and must compromise on one of them? I would agree with that.

Regarding Marty's post, I think the example of comparing the quality of a 35mm image with DV footage is not so relevant. If I compare XL2 and VX2000 4:3 DV footage, I can also see differences in clarity that are obviously not related to resolution.

Even if we talk about sampling 16:9 at 960x480, it is proportionally the same number of horizontal pixels as sampling 4:3 at 720x480. Remember that the XL2 shows a wider view in 16:9 mode, and 1/3 more width than 4:3 requires 1/3 more pixels to maintain the same resolution. So how can we expect any more clarity or higher resolution from a 16:9 image sampled at 960x480 pixels compared with a 4:3 image sampled at 720x480. Even allowing for perfect downsampling (which is another can of worms), I thought the results should be exactly the same (except one is wider).

Richard

An interesting thread to follow. If one of you know the answer, I would be interested to know how the SDX900 acheive to provide the differents aspects ratios. Three 16X9 2/3" with a smaller patch for 4:3, or the same as Xl2 (but bigger CCDs of course)?

Richard: the XL2 offers both native 16:9 *AND* native 4:3

The 16:9 is NOT SAMPLED at 720 x 480, it is STORED at 720 x 480.

That is a WHOLE different thing. The CCD's are sampled at 960 x 480!

That is downsampled to 720 x 480. Again, if you want to strictly
talk resolution the INCREASE is NOT in the HORIZONTAL BUT in
the VERTICAL. You get 480 lines of (original) pixels instead of 364!

So the sampling IS higher and the end result IS better.

You are mixing horizontal with vertical. And yes, you are right
that the field widens so more information is in the horizontal
(which should perhaps yield a higher compression level).

Your comparison to the VX2000 actually has everything to do with
resolution since the sensors on the XL2 use more pixels (which is
whole other thing to talk about...) and the lens is probably sharper
as well. But that's for another thread <g>